Is it possible to achieve an exact filesize using ffvfw?
Is 2 Pass the way to go?

2-pass is always the way to have a fixed filesize; The problem is that it seems the 2-pass crasheds on ffvfw (I did not test myself).

Hm. I had no problems doing a 2-pass
The second pass matched wanted filesize within 1%

The problem I observed with 2-pass is that the VBV Buffer is 14. That means problems in muxing. When I ran it thru the old version of TMPGEnc that had the VBV Optimizer it still only came out at 40.

We need a tool that can really optimize the VBV buffer.

Racer99

Hi guys,
I already digged a lot trying to find such tool but with no luck...
Look forward to see that we can find a solution to the VBV probl though.
C ya tonite.

__________________
Rui

Silly question:
fixing VBV will avoid the problem of too high and too low bitrate?.
Another one, but I guess the answer will be NO: is there a tool to fix too high bitrates?. I suppose no, and I guess that if we just decrease bitrate at some points, without touching Q, this would decrease quality in those scenes where higher bitrate is needed...
I suppose that you will have to tweak more this codec (I'm not good at this, I just can test and share results with you).

uuuhmmm... I saw sth in the ffvfw 2 pass mode options... there was mentioned a max bitrate of higher than 10000 as far as i can remember

(using Linux atm... so no checking for me)

Yes Razorblade2000: in Two passes-second pass there's an option Max bitrate. Don't know what's for, since I didn't try two passes (some people said that this mode crashes ffvfw, and it would make encoding far sloooow, isn't it?).
Do you think we could test two pass (with the advantage of better file size prediction), but making the proccess faster in a way?.

I've tested two ffvfw two pass, and numbers of "max bitrate" and "file size" does nothig I only tested this on very low bitrates (to test if its work) and although qualit of picture was , with max bitrate set to 1024 and file size also 1024 (I like round numbers ) file size was ~13MB with avrg. bitrate ~1800 and max ~6000. This was lotr1 trailer 720x480.
Still testing ...

Karassi, you have positive results, in normal way?


PS: I'm testing something weird, anybody want to try
1. do prediction by XVID
2. insert pass file to StatsReader
3. "rework" this file to "stats" (ffvfw will eats this , AFAI try)
4. go

I'm still working on this.
Benefits: very fast first pass, and faster (2-3 times) second pass.
I think correia will love this

Only one question, is VBV 14 is realy bad. I know that for KVCD myst be 20(24), but WHY?

__________________
Go for SECAM =)

Because that is the standard !

40 (actually displayed as 20 by bitrate viewer) for VCD, 112 for SVCD, 224 for DVD.

Thx Phil, I know standards (few of them )
But what is "phisical" purpose of VBV? First test with ffvfw and VBV=7 was played on "my" standalone with no problem (problem was, I think, with too high bitrate).

__________________
Go for SECAM =)

You were right Hydeus: I would love it
But you were also right Phil: It doesn't meet the standards
Where is the Milan guy after all We're all tired of :banghead: with ffvfw...
If only he would show up just to point us in the right path...
C ya

__________________
Rui

In layman's term, it basically defines a buffer size, so the mpeg player (software or standalone) know how to handle the bitrate allocation, and don't overflow or underflow.
Now here's the full technical definition:

http://www.bretl.com/mpeghtml/VBV.HTM

-kwag

Thx Kwag. And it's not too long

PS:
Now you talking to me

__________________
Go for SECAM =)